What is "method of scoring(analytic derivatives)"?

[Ching]

I am running a SVAR on Eviews version 7, and the estimation method states on the result table is the "method of scoring(analytic derivatives)".

I tried to search this estimation method online and cannot find any.
Could someone kindly explain how this estimation method work?

Many thanks!

[EViews Gareth]

As per the EViews User Guide:

Code: Select all

A and B are estimated by maximum likelihood, assuming the innovations are multivariate normal. We evaluate the likelihood in terms of unconstrained parameters by substituting out the constraints. The log likelihood is maximized by the method of scoring (with a Marquardt- type diagonal correction—See “Marquardt,” on page 922), where the gradient and expected information matrix are evaluated analytically. See Amisano and Giannini (1997) for the analytic expression of these derivatives.

Method of Scoring is a Newton step method using the Information Matrix (expectation of the second derivatives) to form the Hessian matrix.

See Introduction to The Theory and Practice Of Econometics , Judge et.al , for more details.

[Ching]

I see!Thanks a lot EViews Gareth!